How to use the usfiscaldata skill for U.S. Treasury data

What an agent skill is and what usfiscaldata makes the agent do

An agent skill is a folder of instructions and reference files that an AI agent loads when a task matches the skill's description. The usfiscaldata skill covers the U.S. Treasury Fiscal Data API. Once it is loaded, the agent knows the correct endpoints, filter and sort syntax, pagination, and how to read returned values.

The API covers national debt (Debt to the Penny), Daily Treasury Statements, Monthly Treasury Statements, Treasury securities auctions, interest rates, foreign exchange reporting rates, savings bonds, and government revenue and spending statistics. The agent can write and run short Python or R code to fetch and tabulate these figures. The API needs no API key or account.

Install the usfiscaldata skill

Download the ZIP for the skill from this page. Unzip it into your personal skills folder, ~/.claude/skills/, or into a project's .claude/skills/ folder if you want the skill available only in that project.

In the Claude apps, you can instead upload the skill under Settings → Capabilities → Skills.

The skill's own instructions list one dependency install command for the Python examples. Run it in the environment where the agent will execute code.

• Personal install: ~/.claude/skills/usfiscaldata/
• Project install: .claude/skills/usfiscaldata/ inside your project
• Python requirements: Python 3.10+ with requests and pandas
• R alternative: R with httr and jsonlite for the R examples
• Network access to api.fiscaldata.treasury.gov is required

How to trigger the skill with natural requests

You do not need special syntax. Ask for the Treasury figure you want in plain language, and mention Treasury or federal data so the agent matches the skill. The skill description names the topics it covers: national debt, Daily Treasury Statements, Monthly Treasury Statements, auctions, interest rates, exchange rates, savings bonds, and federal revenue and spending.

Be specific about the period and the output you want. A request that names a date range and a format (table, chart, CSV) gives the agent what it needs to write the query correctly.

• What is the current national debt from Treasury? Show the date it was recorded.
• Pull Treasury exchange rates for Canada and Mexico since January 2024 into a table.
• Get the last 30 days of Treasury Operating Cash Balance and plot it.
• List upcoming Treasury securities auctions and save them as CSV.

Know the key datasets and what they cover

The skill maps each topic to an endpoint under the base URL https://api.fiscaldata.treasury.gov/services/api/fiscal_service. Debt to the Penny is updated daily, the Daily Treasury Statement tables are updated on business days, and Monthly Treasury Statement tables are monthly. Auction datasets are updated as needed.

Endpoint paths change over time, so check each dataset's API Quick Guide on the Treasury site before you depend on a path. The skill's reference files give table-level detail for debt, fiscal statements, interest rates and securities, and savings bonds.

• Debt: Debt to the Penny, Historical Debt Outstanding, Schedules of Federal Debt
• Statements: DTS Operating Cash Balance, DTS Deposits & Withdrawals, Monthly Treasury Statement
• Rates and exchange: Average Interest Rates, Treasury Reporting Rates of Exchange, Interest Expense on Public Debt
• Securities: Auctions Data, Upcoming Auctions, Buybacks
• Savings bonds: I Bonds Interest Rates, Savings Bonds Issues, Redemptions & Maturities

Read the values correctly: strings, millions and exchange rates

Data-row values come back as strings, so the agent converts them with float() or pd.to_datetime(). Null values appear as the string "null". Metadata counts are normal JSON numbers.

DTS amounts are in millions. For the Treasury General Account closing balance, the skill reads the open_today_bal field because close_today_bal is null in current data. Treasury reporting rates are foreign-currency units per USD, so you divide a foreign amount by the rate to get USD. These are government reporting rates, not live quotes.

Interest expense has month and fiscal-year-to-date fields, so never sum FYTD values across months. Auctions use auction_date for event timing, while record_date is the publication date.

Run a first query yourself

To check that the setup works, ask the agent for the current national debt. The skill's quick start requests the latest row of Debt to the Penny, sorted by record_date in descending order, with a page size of 1, and prints the tot_pub_debt_out_amt value.

The API has no key, so a failure is usually a network or path problem. Check that the sandbox or machine can reach api.fiscaldata.treasury.gov, then confirm the endpoint on the dataset's API Quick Guide.

• Parameters: fields=, filter=, sort=, format= (json, csv or xml), page[size]= (default 100), page[number]= (starts at 1)
• Filter operators: lt, lte, gt, gte, eq, in
• Multiple filters are separated by commas

Safety notes: scripts and network access

This skill runs scripts and uses the network. The agent executes Python or R code that calls api.fiscaldata.treasury.gov, and its declared tools include Read, Write, Edit and Bash. Review the code the agent writes before running it, especially if you run it on a machine with sensitive files.

The skill needs no credentials. Do not paste API keys or passwords into prompts for this skill, because none are used. The API is open and free for commercial and non-commercial use, according to the reference notes.

• Network: only api.fiscaldata.treasury.gov is needed for these queries
• Tools declared: Read, Write, Edit, Bash
• No credentials are required